Learnings

Outcomes

Two detailed reports were prepared by UoM which provide several recommendations for the ESO. The reports were released in January 2021 and are published on the Smarter Networks Portal (see link above). The extension of the project enabled the development of a new tool to perform the LWWR analysis. This tool has since been embedded in the business; included in the NOA process as well as being utilized throughout the year in various CBAs.

Recommendations from the 1st report

In terms of possible improvements to the current NOA process and more general aspects that might be worth exploring, the following are recommended: 

• Additional operational snapshots could be included in the technical analysis besides the current winter peak assessment. This is in light of increasing operational complexity and uncertainty that might drive worst case flows across boundaries at different times of the year. Similarly, inclusion of new operational characteristics and constraints, such as associated with low-inertia conditions, could be desirable.
• The use of LWR or similar approaches could be adopted to inform the optimality of interconnector assessment across scenarios too. This would allow a more consistent and integrated approach between selection of (internal) boundary reinforcement and optimal level of interconnectors.
• A Least Worst Weighted Regret (LWWR) approach, in which scenarios have explicitly assigned weights could be proposed as the generalised version of LWR (where equiprobable scenarios are implicitly assumed).
• The implicit equiprobable scenario representation of the current LWR approach can be interpreted as a special case of the more general LWWR whereby it is intrinsically believed that all current scenarios are similarly plausible and likely to happen28. However, if there are reasons to consider asymmetry in the likelihood of the considered scenarios, consideration might be given to a more detailed assessment that might explore different probability weights.
• While we are aware that the selection of probability weights to assign to scenarios may be difficult and controversial, our proposed unified framework provides a consistent and comprehensive view that could seamlessly compare and assess the outcomes of (apparently) different methodologies (i.e., probabilistic, LWWR and min-max weighted cost), with scenario weights being considered as a natural component. This could eventually provide more transparency and robustness to the investment process and the selected options, thus resulting in reduced risk of spurious solutions and reduced risk of decisions being driven more by specific scenarios than methodologies, and in general enhanced hedge against potential uncertainty. Furthermore, by modulating the value of the scenario weights, the impact of different degrees of risk-aversion may also be explored.
• Such analysis could also be supported by visual tools that could identify decision-stability regions with win-win solutions from different methodologies and suggest what solutions might require further analysis, for example in terms of expected costs and regrets.
• Irrespective of the formal use of probability weights, multi-parametric scenario sensitivity studies could be performed to provide insights into the benefits and risks of different proposed solutions under different possible future occurrences, with for example expected costs and expected regrets (a measure of risk) analysis for different methodologies able to clearly assess all the implications of using different approaches, risk-aversion degrees, and scenario weights. 

Recommendations from the 2nd report:

• The uncertainty brought by increasing penetration of renewable energy might require network planners to redefine their planning methodology, so that it can account for the impact of this uncertainty in the evaluation of network’s technical performance, and consequently accommodate these changes in the cost-benefit analysis which decides the worthiness of a network investment. NGESO has already been actively developing a probabilistic analysis to enhance the existing deterministic one which only derives boundary capability in winter peak snapshot. We have made recommendations which can enhance the current methodology from several perspectives, such as using different sampling techniques to derive boundary capability setpoints, proposing reliability indices and risk metrics to monitoring network performance, and using machine learning techniques to perform network security assessment and predict boundary capability contribution from reinforcement options.
• Although the computational time of NOA’s CBA could be substantially reduced by using static37 boundaries to represent network constraints in system operational modelling, with more and more variable power flows it may introduce more inaccuracy in measuring constraint cost reduction contributed by reinforcement options. Therefore, it may be beneficial to integrate the network technical modelling into the CBA’s economic dispatch analysis; this would consequently enable a more precise calculation of network constraint costs in NOA’s CBA.
• Commercial solutions can be a great complement to network-based options for network reinforcement, as these solutions feature significant technical and economic flexibility. Taking grid-scale battery as an example, it might be able to provide different ancillary services (e.g., frequency response, balancing mechanism) besides reducing network congestions. Commercial solutions have a more flexible service contract length (e.g., 5-10 years), which could be shorter than the lifetime of network-based options (e.g., 30-40 years from transmission lines). This feature represents a valuable factor in investment flexibility. NGESO has also been developing commercial solutions which are included in the reinforcement options list published in NOA. However, the methodology of commercial solutions evaluation, which is explained in 2.3, could be further improved by integrating it into the NOA framework to evaluate the constraint costs under a unified framework. 

Regarding the current methodology of NOA’s CBA, it was analysed in detail in the first report, focusing on discussing the core elements of the decision-making process. In this report we delved deeper in the aspects related to the identification of investment options that would provide more flexibility across scenarios: 

• The structure of decisions associated with each reinforcement option seems to provide sufficient flexibility to initiate, hold or stop a project. However, this structure of decisions is used in the context of a two-stage decision process where many of the decisions are fixed based on the deterministic assessment of the scenarios. This is reducing the space of investment strategies that can be selected by the LWR to a point where possibly very little flexibility can be captured.
• Seeking more flexibility is a process that in general has very high computational requirements because many combinations of options have to be assessed under different operation conditions; in this context, in the short-term, we recommend to focus in automatising the deterministic assessment of each scenario as much as possible, reduce computational burden by finding the right periods of operation of the system that capture the system behaviour for each year and each scenario, and improving the selection of the reinforcements that enter the LWR in order to be able to capture more flexibility.
• In the long run the CBA assessment should aim to evolve into a multistage integrated model that can evaluate network and non-network solutions without the need to first assess the operation of the system for all combinations of investment options in all scenarios. These integrated models are currently available for risk-constrained stochastic approaches; however, no efficient solutions exist, to our knowledge, in the realm of multistage LWR. Not evolving into an integrated approach in the long run might render the current methodology inadequate to fully assess flexible investment options and incorporate more complex representations of future decisions; this, in turn, might generate larger regrets than those that the current methodology is capable to prevent. 

Lessons Learnt

The report provided by UoM is a review by an independent party of the NOA methodology. It shows an outside perspective of the ESO’s approach and compares it to that adopted by other countries. The report contains recommendations on new tools and ideas to perform the NOA CBA and technical probabilistic analysis and includes roadmaps to implement them. Furthermore, the report provides a very detailed description of ESO methods and will be a useful resource for interested readers to better understand the work the NGESO undertakes.  

More generally, the learnings demonstrate that there are several ways the NOA process can be improved, these vary in scale and ease of application. Some are implementation-ready, such as the LWWR technique and tool, while others will require further consideration, research or follow on projects.